Auto-pick first breaks with complex raypaths for undulate surface conditions

Abstract

First-break picking is the key step in seismic data processing for surveying undulate surfaces, and directly influences the precision of near-surface modeling and effects of static corrections. The current first-break auto-picking methods may fail when the signal-to-noise ratio (SNR) is low for seismic data in the undulate area, and require labor and time intensive manual picking. This study develops an improved super-virtual interferometry (SVI) method that combines multichannel and multidomain quality control (MMQC) techniques to achieve auto-picked first breaks. The improved SVI method extends the SVI application to enhance the SNR for near-surface scattered waves for the first time, which allows for the SVI method to adapt to first breaks with complex raypaths by linear combination of refractions and near-surface scattered waves. Methods of inverse and multidomain interferometry are developed to effectively enhance the virtual records extracted by the SVI method. The deconvolution filter for waveforms is used to increase resolution and reduce false picks, while the MMQC technique is designed to auto-correct false picks and increase the stability of auto-picking first breaks. The robust technique developed in this study enables stable processing of large 3D seismic datasets. Higher quality results are obtained using the approach presented in this paper to actual field data from the mountain areas in western China, when compared to some commonly used commercial software.